Address Validation and Address Verification: What They Are and How They Differ
Razvan Simion, Lead Software Engineer

Address Validation and Address Verification: What They Are and How They Differ
Three terms get used as if they were one: validation, verification and standardisation. They describe different checks with different costs and different failure modes, and choosing the wrong one is how teams end up paying for a service that does not solve their problem.
This article separates them, then looks at what an address validator actually does, where bad addresses come from, and when to check addresses as they arrive rather than in a nightly batch.
Address validation, verification and standardisation: the actual differences
Address validation asks whether an address is valid: do these components exist, and do they belong together? A postcode that does not match the street it was typed with fails validation immediately. So does a house number that does not exist on that street.
Address verification goes one step further and confirms the address is real and reachable, usually by matching it against an authoritative source such as a postal operator's delivery file. A validated address is well formed. A verified address is one someone can actually deliver to.
Address standardisation is the formatting layer. It takes "flat 2, 14 high st." and returns "Flat 2, 14 High Street" with each component in its own field, cased and abbreviated consistently. It makes no claim about whether the address exists; it only makes your data uniform enough to compare, deduplicate and print.
In practice you want all three, and most services do all three in one call. The distinction still matters when you are reading a contract or a response payload: "validated" and "verified" are not the same guarantee.
What an address validator checks
A serious address validator runs several checks, not one:
- Component existence: does this street exist in this town, does this postcode exist at all.
- Cross-field consistency: do the postcode, street, number and town describe the same place.
- Premise level: does the building number, and the flat or unit within it, exist.
- Deliverability: is this a delivery point, or is it an address that exists but receives no post.
- Normalisation: is the result returned in a consistent, structured form.
The difference between this and a required-field check on a form is total. A required-field check accepts "SW1A 1AA" paired with any street at all. Real validation rejects the combination.
Where bad addresses come from
Almost all of them come from typing. A customer transposes two digits in a house number, autocompletes to the wrong town, or abbreviates a street name in a way your system does not recognise. A smaller share comes from data ageing: new-build properties that did not exist when your list was compiled, streets renamed, or postcodes reallocated.
Both problems have the same answer, applied at different points. Typing errors are best caught at the moment of entry, while the customer is still there to correct them. Ageing is best caught by a periodic pass over stored data. Neither substitutes for the other.
Real-time validation vs bulk address validation
Real-time validation happens in the form. The customer types, the API confirms, and the record is clean before it is ever stored. This is where the money is: a corrected address costs nothing, while a failed delivery costs postage, handling time and a disappointed customer.
Bulk validation runs over a list you already have. It is how you find out how much of your existing database is wrong, and it is usually a sobering number. Run it once to size the problem, fix what is fixable automatically, and flag the rest for review rather than silently rewriting customer records.
The two are complementary. Real time stops the leak; bulk cleans up what has already leaked. If you only do one, do real time, because the bulk problem stops growing the moment you do.
What it costs you when an address is wrong
The direct cost is the re-delivery: postage twice, handling twice, sometimes a courier surcharge for a returned parcel. The indirect costs are larger and less visible.
A failed delivery generates a support contact, and support contacts are expensive. It delays revenue recognition on the order. If it happens at the wrong moment it produces a refund and a lost customer rather than a re-send. And in aggregate, addresses that fail silently pollute every downstream analysis you run on where your customers are.
Choosing an address validation tool
Ask five questions of any provider:
- What is the underlying data source, and how often is it refreshed? For the UK this usually means Royal Mail PAF or Ordnance Survey AddressBase. For the Netherlands it means the BAG. "Proprietary database" without further detail is a warning sign.
- What coverage do you actually need? A UK-only tool is fine until your first order from Dublin or Berlin. If that is on the horizon, look at international addresses before you commit.
- What happens to a partial match? The useful answer is a suggestion the customer can accept. The unhelpful answer is a hard rejection.
- How fast is it? Anything that noticeably delays a checkout is costing you more than it saves.
- What does it return, not just what does it check? Structured components and coordinates are worth far more than a boolean.
If you are past evaluation and into implementation, the developer view of the same job covers the request and response shape in detail. Validation is one piece of a larger set, and our guide to UK address and postcode APIs places it alongside lookup, autocomplete and geocoding.
Frequently asked questions
What is the difference between address validation and address verification?
Validation checks that an address is well formed and internally consistent: the postcode, street and number describe a real place that fits together. Verification confirms against an authoritative source that the address exists and can receive post. Most commercial services do both and describe it as one product.
Is there a free address validation tool?
There are free single-address checkers, and several providers offer a free development tier of their API. Free bulk validation of a large customer database is rare, because the underlying address data is licensed.
How do I validate an address in bulk?
Export the addresses, run them through the provider's batch endpoint or file upload, and split the results three ways: matched exactly, matched with a correction, and unmatched. Apply the exact matches automatically, review the corrections, and investigate the unmatched rather than deleting them. Many will be new-build properties.
Can address validation reject a real address?
Yes. New-build properties, recently renamed streets and unusual sub-premises can all fail against a data file that has not caught up. This is why a good validation flow lets the customer proceed with an unmatched address after confirming it, rather than blocking them.
Does validating addresses improve delivery rates?
It removes one specific cause of failure, which is an address that was never correct. It does nothing about a customer who is out when the courier arrives. Expect a meaningful reduction in returns, not a perfect delivery rate.
